ENVS 291 Transition to R course will be offered in Winter Quarter 2015.

UCSC graduate students can sign up, as space allows, for ENVS 291-03 (43166)(G Gilbert).

Meetings are TTh 2:00-3:45 Jan 8-28 for the core of the course;
Remaining meetings are 19 Feb, 26 Feb, and 5 March for classes 8, 9, 10.

The course will not be offered again until 2017.

Class 9 - More Advanced Graphics.docx
Note:must install ggplot2 and grid packages

shows currently availble objects in workspace

getDropboxCSV<-function(){
cat("From Dropbox in your browser, locate the desired .csv file, \nclick Share, and then copy the URL \nfrom the Link to File dialog box.\n")
cat("Paste the URL for your .csv file here: ")
myname<-readLines(n=1)
mynamefix<-sub("www.dropbox","dl.dropboxusercontent",myname)
mynamefix<-sub("dl=0","",mynamefix)
mynamefix<-sub("\\?","",mynamefix)
bin<-getBinaryURL(mynamefix,ssl.verifypeer=FALSE)
justname<-unlist(strsplit(mynamefix,split="/"))[6]
con <- file(justname, open = "wb")
writeBin(bin, con)
close(con)
read.csv(justname)
}

bbb<-getDropboxCSV()

Defining the pathway to load a file:
m<-read.csv(file="c://marm/teaching/rclass/mydataset.csv") #note // after c:
or
m<-read.csv(file="/marm/teaching/rclass/mydataset.csv") #note leave out c:

To scroll through the graphic window, on a mac just use the left and right arrows.
On a PC:
1. Create a graphic.
2. With the graphic window active, go to the "History" tab, next to "File."
3. Click on "Recording." Make sure it is checked.
4. Now each graph you make from this time forward in the current R session will be recorded. You can scroll through graphs using "PgUp" and "PgDn.

For better R editors for the PC, try NppToR and Notepad++
http://sourceforge.net/projects/npptor/
http://notepad-plus-plus.org/

Line and arrow codes
x1=rep(1,6); x2<-rep(3,6); y<-seq(6,1); linecodes<-seq(1:6)
plot(0,0,xlim=c(0,10),ylim=c(0,6.2),pch=1,col=0,axes=FALSE,xlab="",ylab="")
for(i in 1:6){lines(c(x1[i],x2[i]),c(y[i],y[i]),lty=linecode[i])}
text(x1-.8,y,linecode,pos=4); text(1.5,0.1,"lines\nlty",pos=4)

for(i in 1:6){lines(c(x1[i]+3,x2[i]+3),c(y[i],y[i]),lty=linecode[i],lwd=linecode[i])}
text(x1-.8+3,y,linecode,pos=4); text(4.5,0.1,"lines\nlwd",pos=4)

for(i in 1:3){arrows(x1[i]+6, y[i],x2[i]+6, y[i],code=linecode[i])}
text(x1[1:3]-.8+6,y[1:3],linecode[1:3],pos=4); text(7,0.1,"arrow\ncode",pos=4)

Color by numbers

x<-rep(seq(1,3,1),3);y<-sort(x);z<-seq(0,8,1)
plot(y~x,pch=15,cex=5,col=z,xlim=c(0,4),ylim=c(0,4),axes=F,xlab="",ylab=""); box()
text(y~x,labels=c(0:8),col=c(1,8,rep(1,7)))

Color by names

colorlist<-read.csv("http://people.ucsc.edu/~ggilbert/Rclass_docs/colorlist.csv")
rect<-as.matrix(cbind(rep(1,580),rep(1,580)))
y<-rep(seq(1,58,1),10); x<-sort(rep(seq(1,10,1),58))
z<-as.character(colorlist$color);textcol<-colorlist$textcode
symbols(y~x,rectangles=rect,xlab="",ylab="",bg=z,xlim=c(0.5,10.5),ylim=c(0,59),inches=FALSE); box()
text(y~x,labels=z,col=textcol,cex=.5)

Click on image for large version

Here is a like to a good web site from HTML Color Codes to get, well, HTML color codes
http://html-color-codes.info/

You can then call the color you want as:

plot(y~x,pch=15, col="#8E2BC3")

or use rgb(red,green,blue,alpha) to control the RGB colors and intensity
mycol1<-rgb(142/255,43/255,195/255,.7) #the values here all go from 0 to 1 where max BCG is 255
plot(y~x,pch=19,col=mycol1)

These are the data used for all the scatterplot examples

x<-c(1,2,3,4,5,6,7,8)
y1<-c(2,4,5,7,8,7,9,10)
y2<-c(1,3,2,4,6,5,7,7)

#Basic scatterplot, adding labels (xlab and ylab) and setting min and max of axes (xlim,ylim)

# Overlay a second series of y values, control symbol with pch and color with col,  and add a legend.

plot(x,y1,xlab="arrival order",ylab="hat size (cm)",ylim=c(0,10),xlim=c(0,8),pch=1,col="black")

points(x,y2,pch=19,col="blue")

legend(0,10,c("male","female"),pch=c(1,19),col=c("black","blue"))

#Connect the points with lines
#type"p"=points, "l"=lines, "b"= both,"c" lines alone of "b",
#"o" =overplotted,"h"=histogram-like,"s"=stair steps

plot(x,y1,xlab="arrival order",ylab="hat size (cm)",ylim=c(0,10),xlim=c(0,8),pch=1,col="black",type="b")

#just lines, controlling line thickness with lwd and line type with lty
#add a legend in the upper left

plot(x,y1,xlab="arrival order",ylab="hat size (cm)",ylim=c(0,10),xlim=c(0,8),pch=1,col="black",type="l",lwd=2,lty=2)

lines(x,y2,pch=19,col="blue",type="l",lwd=1,lty=1)

#add smooth lowess curves to  each set of points in the scatterplot

plot(x,y1,xlab="arrival order",ylab="hat size (cm)",ylim=c(0,10),xlim=c(0,8),col="dark green",pch=1,lwd=2)
lines(lowess(x,y1),lwd=2,lty=3,col="dark green")
points(x,y2,pch=19,col="dark blue")
lines(lowess(x,y2),lwd=2,lty=2,col="dark blue")

#Use abline to add linear regression lines to each set of points in the scatterplot

plot(x,y1,xlab="arrival order",ylab="hat size (cm)",ylim=c(0,10),xlim=c(0,8),col="black",pch=1,lwd=1)
abline(lm(y1~x),lwd=1,lty=1,col="black")

points(x,y2,pch=19,col="blue")
abline(lm(y2~x),lwd=1,lty=2,col="blue")

#get the relevant statistics for the regression line, then put on the graph as text
a<-summary(lm(y2~x))  #this puts summary stats of the linear regression of y2 on x into list a
R2<-signif(a$adj.r.squared,3)  #adjusted R squared
F<-signif(a$fstatistic[1],3) #F statistic
ndf<-signif(a$fstatistic[2],1)  #degrees of freedom numerator
ddf<- signif(a$fstatistic[3],1)  #degress of freedom denominator
P<-signif(a$coefficients[2,4],4)   #P value for significant slope

plot(x,y2,xlab="arrival order",ylab="hat size (cm)",ylim=c(0,10),xlim=c(0,8),col="blue",pch=19,lwd=1)

abline(lm(y2~x),lwd=1,lty=1,col="blue")  #puts in the regression line

text(0,9,paste("F=",F,", df=",ndf,",",ddf,"\n","R^2=",R2,", P=",P,sep=""),pos=4)  #adds the statistics

#create a vector of 100 random numbers from a normal distribution, with mean=32 and sd=4.1
> data<-rnorm(100,mean=32,sd=4.1)

#Use hist to make histograms setting bins by range and size
hist(x=data, breaks=seq(from=24,to=44, by=2), freq=TRUE, col="blue", xlab="Ca (ppm)", ylab="Number of samples", main=NA)

#Use hist to make histograms setting bins by number of bins
hist(x=data, breaks=15, freq=TRUE, col="orange", xlab="Ca (ppm)", ylab="Number of samples", main=NA)

#Use hist to make histograms with proportions rather than counts
hist(x=data, breaks=15, freq=TRUE, col="orange", xlab="Ca (ppm)", ylab="Number of samples", main=NA)

# freq = TRUE gives counts

#Use Lattice histogram to Make Histograms setting bin size
#Short version:
library(lattice)
histogram(x=data, col="blue", type="count", xlab="Ca (ppm)", ylab="Number of samples", breaks=seq(from=20,to=45,by=2), aspect=.75, scales=list(cex=1.2,tck=c(1,-1)))

#Annotated version
histogram(
x=data, #data column from which to make the histogram
col="blue", #fill color for columns see colors() for list
type="count", #sets the vertical axis - count, percent, or density
xlab="Ca (ppm)", #label for x axis
ylab="Number of samples", #label for y axis
breaks=seq(from=0,to=5,by=.5), #min, max, and bin interval for bins
aspect=.75, #ratio of vertical to horizontal plot shape desired
scales=list(cex=1.2,tck=c(1,-1))) #sets tick marks and numbers on axes
#cex is the multiple of the default size for numbers font
#tck=c(left/bottom,right/top) tick size (negative for inside)

#Make a Quantile-Quantile Plot to examine normality of distribution

Basic barplot of mean values for each treatment.

barplot(myData$mean, xlab="Treatments", ylab="Plant biomass (g)", names=myData$treat, col="grey", ylim=c(0,10)) 
#gives basic barplot of means

box()   #draw a box around the plot to make it pretty

Add error bars
bp<-barplot(myData$mean)  #Gets the x-midpoints of each bar
# they serve as x values for error bars, here they are 0.7, 1.9, 3.1, 4.3

barplot(myData$mean, xlab="Treatments", ylab="Plant biomass (g)", names=myData$treat, col="grey", ylim=c(0,10)) 
#gives basic barplot of means
box()   #draw a box around the plot to make it pretty

# add ± 1 s.d. lines to each mean
arrows(bp, myData$mean-myData$sd, bp, myData$mean+myData$sd, lwd=1.5, angle=90, length=0.1, code=3)

#code =3 is double headed, angle=90 is flat heads, length=0.1 is width of head

Turn it horizontal
bp<-barplot(myData$mean)

barplot(myData$mean, xlab="Plant biomass (g)", ylab=" Treatments ", names=myData$treat, col="grey", xlim=c(0,10), horiz=TRUE)
box()
arrows(myData$mean-myData$sd, bp, myData$mean+myData$sd, bp,lwd=1.5, angle=90, length=0.1, code=3)

Add designation of significant differences among bars

bp<-barplot(myData$mean)
barplot(myData$mean, xlab="Treatments", ylab="Plant biomass (g)", names=myData$treat, col="grey", ylim=c(0,11))
box()
arrows(bp, myData$mean-myData$sd, bp, myData$mean+myData$sd, lwd=1.5, angle=90, length=0.1, code=3)
text(x=(bp-.2),y=myData$mean+.4,labels=c("A","B","B","A"))
text(x=4.3,y=9, labels="F=5.7\ndf=3,15\nP≤0.03")

#use text to add whatever labels to the plot

data1   If your data look like this, you must first reshape them
Sex Species count
Male       A     5
Male       B     7
Male       C     9
Male       D    10
Female    A     4
Female    B     6
Female    C    10
Female    D    11

Reshape the data at left to a table format:
re<-reshape(data1,idvar="Sex",timevar="Species",direction="wide")
names(re)<-c("Sex","A","B","C","D")
rem<-as.matrix(re[,2:5])   #the counts data need to be in a matrix form

re                                               rem
Sex      A B  C   D                     A  B   C   D
Male    5  7   9  10                     5   7    9  10
Female 4  6 10  11                     4   6  10  11

Stack the male and female bars within a species
barplot(rem,xlab="Species",ylab="Number of individuals", col=c("light blue","green"),beside=FALSE)
box()  #puts a pretty box around it

legend(0.2,20,re$Sex,bty="o",fill=c("light blue","green"))

Put the males and females size by side
barplot(rem,xlab="Species",ylab="Number of individuals",col=c("light blue","green"),beside=TRUE)
box()

Data frame "d" for these pie charts includes
group count
    a       40
    b       20
    c       30
    d       10

pie(d$count,labels=d$group)

Pretty up the labels:

Change the colors and rotate the pie

Make it cubist to reduce usefulness and readability

library(lattice) # load the lattice package
Create data frame d using code at right

# simple xy scatterplot, putting the y variable on a log 10 scale (log="e") for natural log
# ticks are outward on Left/Bottom and inward for Right/Top
#pch=19 uses filled circles

xyplot(y1~x1, data=d, pch=19, scales=list(tck=c(1,-1), y=list(log=10)))

# Make a plot with different symbols for males and females (variable sex)

xyplot(y1~x1, data=d, group=sex,
par.settings=simpleTheme(cex=1.2,pch=c(19,1),col=c("blue","red")), #uses simpleTheme to set symbol attributes
auto.key=list(TRUE,x=.02,y=.9), #use the par.settings to create a legend in the upper left. x and y are 0-1 scale
scales=list(tck=c(1,-1))) #set where and direction of ticks

#Make split scatterplot putting each group (sex) into a separate grid panel

xyplot(y1~x1 |sex, data=d,
pch=19, scales=list(tck=c(1,-1),alternating=c(1,1)))
#tck c(1,-1) sends left/bottom ticks out and right/top ticks in
#alternating sets labels bottom/left for both panels (0=none, 1=B/L, 2=T/R, 3=Both)

#Make 2x2 grid split scatterplot with multiple grouping variables (sex and age)

xyplot(y1~x1 |sex+age, data=d,
pch=19, scales=list(tck=c(1,-1),alternating=c(1,1)))

#tck c(1,-1) sends left/bottom ticks out and right/top ticks in

#alternating sets labels bottom/left for both panels (0=none, 1=B/L, 2=T/R, 3=Both)

#plot two dependent variables in the same grid panel, split on sex

xyplot(y1+x2~x1 |sex, data=d,
pch=c(19,1), scales=list(tck=c(1,-1),alternating=c(1,1)),
par.settings=simpleTheme(cex=0.9,pch=c(19,1),col=c("blue","red")),
auto.key=list(TRUE,x=0.01,y=0.9),ylab="y1 or x2")

#tck c(1,-1) sends left/bottom ticks out and right/top ticks in
#alternating sets labels bottom/left for both panels (0=none, 1=B/L, 2=T/R, 3=Both)
#par.settings uses simpleTheme function to set symbol attributes
#auto.key places a key in upper left

#split xyplots into two grid panels based on one factor (sex)
#and within a panel, use different symbols for a second factor (age)

xyplot(y1~x1 |sex, groups=age, data=d,
pch=c(19,1), scales=list(tck=c(1,-1),alternating=c(1,1)),
par.settings=simpleTheme(cex=0.9,pch=c(19,1),col=c("blue","red")),
auto.key=list(TRUE,x=0.01,y=0.9))

#make a density plot of variable y1, including outward tick markson Left/Bottom but none on Right/Top

densityplot(~y1,data=d,scales=list(tck=c(1,0)))

# dotplot is used for plotting points of continuous variables, grouped by factors

dotplot(y1~sex, data=d, pch=19,
col=c("red","blue"), scales=list(tck=c(1,-1)))

#box and whiskers plot using bwplot
#box is 25,50, 75 quartiles and whiskers are min and max

bwplot(y1~sex, data=d,
col=c("red","blue"), scales=list(tck=c(1,-1)))

#box and whiskers plot using bwplot, split into grid panels based on age
#box is 25,50, 75 quartiles and whiskers are min and max

bwplot(y1~sex | age, data=d,
+ col=c("red","blue"), scales=list(tck=c(1,-1)))

x<-rep(seq(1,10,1),10); y<-sort(x); z<-rnorm(100,mean=10,sd=6) #create 3-D data

levelplot(z~y*x)

x<-rep(seq(1,10,1),10); y<-sort(x); z<-rnorm(100,mean=10,sd=6) #create 3-D data

contourplot(z~y*x)

x<-rep(seq(1,10,1),10); y<-sort(x); z<-rnorm(100,mean=10,sd=6) #create 3-D data

wireframe(z~y*x)